David Sifry

Dave is just this guy, see? Founder of Offbeat Guides, Technorati, Linuxcare, others. Exec at Lyft, Reddit. Currently VP Center on Technology and Society, ADL.

[David Sifry @dsifry]

Most vibe coders don't know what branch protection is. Here's why it matters:

Without it, anyone (including your AI tool) can push directly to main. No review. No checks. One hallucinated refactor and your production app is down.

It takes 2 minutes to turn on in GitHub settings

[David Sifry @dsifry]

I've been shipping production systems for 30 years. The pattern is always the same: something breaks at 2am, you dig in, and you find the gap that was there all along.

I built RepoFortify because I got tired of finding these gaps after the incident. Now I find them before.

[David Sifry @dsifry]

Honest question for anyone building with Cursor, Bolt, or Claude Code:

If your AI tool silently deleted a file it generated last week, would you notice?

[David Sifry @dsifry]

The workflow I wish I had 10 years ago:

1. Push code
2. RepoFortify scans it automatically across 9 dimensions
3. Open Cursor / Claude Code
4. AI reads the fix plan directly via MCP
5. AI fixes the issues
6. Push again. Score goes up.

No dashboards to check. No reports to read

[David Sifry @dsifry]

Watched a founder go from a 31 to a 74 in one afternoon. Didn't write a single line of code.

Connected RepoFortify's MCP server to Cursor, asked for a fix plan, and let Cursor execute it.

Dead exports cleaned up. Missing tests generated. CI enforced.

[David Sifry @dsifry]

Unpopular opinion: AI coding tools are creating a generation of founders who ship faster than any human ever has — and have absolutely no idea how fragile what they've built actually is.

The bottleneck was never writing code. It was knowing what to check after you wrote it.

[David Sifry @dsifry]

Scanned a repo yesterday. Landing page looked great. Stripe integration worked.

Score: 23/100.

Zero tests. No branch protection. 14 dead files. One bad merge away from charging customers twice.

"It works" is not the same as "it's ready."

[David Sifry @dsifry]

@AlexFinn This is a solid workflow. One thing I'd add: after the build, scan the repo. Most AI-gen code ships with no branch protection, missing tests, and stale deps. I built https://t.co/FKf408VsQ5 for exactly that - 9-dimension scan, score out of 100, w/MCP server so yr AI can fix

[David Sifry @dsifry]

@karpathy Have a look at https://t.co/KUz4doNgb8 - I built it to use AI as an internet research assistant, using the best of LLMs and the best of search together to give you topical, referenced research assistance on any topic.

[Yael Eisenstat @YaelEisenstat]

My team @ADL Center for Tech & Society is hiring! Excited to share we have 3 open positions to drive impact in the global fight against online hate & harassment. We're fierce advocates for making digital spaces safe, respectful and equitable for all people. Share far & wide. 🧵1/

[David Sifry @dsifry]

Ever wanted to build your own hedge fund and run algorithmic models with your own trading account? Super excited that one of my angel investments, https://t.co/MbE1vMSud4 does just this, and is in early access beta. You can sign up here: https://t.co/ixIGpXBr3q

[David Sifry @dsifry]

@doctorow @rgibli @doctorow is always a must-read. Agree or disagree with him, he is so often dead on about the near future that you’d be a fool to dismiss him.

[David Sifry @dsifry]

@gabemac The inimitable @gabemac posts a powerful and touching example of social connection and friendship helping a friend come through deep depression. The “dead man’s switch” (tracking loss of signal) is a powerful pattern when thoughtfully applied.

[Christina @cwodtke]

If you hate OKRs, someone is doing them wrong. Maybe you, maybe your boss.

[Jewish Insider @jewishinsider]

“Self-regulation is clearly not working,” @ADL 's @dsifry said at the hearing. “Without regulation and reform, they will continue to focus on generating record profits at the expense of our safety and the security of our republic.”

@marcrod97 reports.

https://t.co/q9ygOp2vlq

[Homeland Security & Govt. Affairs Committee — Dems @HSGAC]

Reports indicate Facebook knew their platforms spread dangerous content. WATCH Chairman @SenGaryPeters ask @ADL’s @Dsifry about the apparent conflict between social media companies’ business models that increase engagement and efforts to take down extremist content.

[NYU's Center for Social Media, AI, and Politics @CSMAP_NYU]

Misinformation spreads rapidly online. In response, Facebook & @Twitter have suggested using ordinary users as fact checkers. But our new #OpenAccess article, in the inaugural issue of @journalsafetech, finds this is likely not a viable solution 1/ https://t.co/IMWFPR76FK

[Tom Coates @tomcoates]

My new cross street. I ❤️ San Francisco.

[грешник @bazengang]

🤟

[Jeff Richards @jrichlive]

This is awesome